The application of the artificial neural network and taguchi method to process sequence design in metal forming processes

This study shows the correlation between the design methodology of the artificial neural network (ANN) and the statistical design of experiments (DOE) approach and is demonstrated for process design in various metal forming processes. After investigating the effect of each parameter upon the characteristics by the Taguchi method which is one of the DOE, orthogonal array (OA) table and characteristics are applied to ANN as experimental data and then opiimal design parameters are established. Using the rigid plastic FEM, the simulations are performed and the results of ANN are confirmed. This technique requires smaller runs than the conventional method to find the optimal condition of design parameters for the design’s aim. This new technique can be used in a wide range of metal forming process designs.     

Application of an optimized SA-ANN hybrid model for parametric modeling and optimization of LASOX cutting of mild steel

Laser assisted oxygen cutting (LASOX) process is an efficient method for cutting thick mild steel plates compared to conventional laser cutting process. However, scanty information is available as to modeling of the process. The paper presents an optimized SA-ANN model of artificial neural network (ANN) and simulated annealing (SA) to predict and optimize cutting quality of LASOX cutting process of mild steel plates. Optimization of SA-ANN parameters is carried out first where the ANN architecture and initial temperature for SA are optimized. The optimized ANN architecture is further trained using single hidden layer back propagation neural network (BPNN) with Bayesian regularization (BR). The trained ANN is then used to evaluate the objective function during optimization with SA. Experimental dataset employed for the purpose consists of input cutting parameters comprising laser power, cutting speed, gas pressure and stand-off distance while the resulting cutting quality is represented by heat affected zone (HAZ) width, kerf width and surface roughness. Results indicate that the SA-ANN model can predict the optimized output with reasonably good accuracy (around 3%). The proposed approach can be extended for prediction and optimization of operational parameters with reasonable accuracy for any experimental dataset.     

Data acquisition,signal processing and modelling: a study of a conducting polypyrrole formate biosensor Part 1: Batch experiment

In this paper a new computerized approach to detection of trace concentrations of formate is considered. In this approach an integrated artificial neural network (ANN) /conducting polymer biosensor is designed. The data collected (current signals) from amperometric detection of the polypyrrole formate biosensor were transferred into a MATLAB^™ environment for data and signal processing. Later the data were transferred into an ANN trained computer for modelling and prediction of output. Such an integrated ANN/polypyrrole biosensor system is capable of prediction of formate concentration based on the created models and patterns. The method has advantages of being more selective and more accurate over conventional methods.     

Hybrid modelling of 7449-T7 aluminium alloy friction stir welded joints

Abstract

Many finite element models use adjustable parameters that control the heat loss to the backing bar, as well as the heat input to the weld. In this paper, we describe a method for determining these parameters with a hybrid artificial neural network (ANN) coupled thermal flow process model of the friction stir welding process. The method successfully determined temperature dependent boundary condition parameters for a series of friction stir welds in 3·2 mm thick 7449 aluminium alloy. The success of the technique depended on the method used to input thermal data into the ANN and the ANN topology. Using this technique to obtain the adjustable parameters of a model is more efficient than the conventional trial and error approach, especially where complex boundary conditions are implemented.     

Artificial Neural Network Prediction of Fretting Wear Behavior of Structural Steel,En 24 Against Bearing Steel,En 31

In this study, artificial neural network (ANN) technique is used to predict the friction and wear behavior of various surface-treated structural steel (En 24) fretted against bearing steel (En 31). A three-layer neural network with a back propagation algorithm is used to train the network. Fretting wear volume and coefficient of friction obtained at different normal loads (ranging between 2.4 and 29.4 N) for various treated samples (hardened, thermo-chemically treated, MoS₂ coated) were used in the formation of training data of ANN. Results of the predictions of ANN are in good agreement with the experimental results. The degree of accuracy of predictions was 96.3 and 95.7% for fretting friction coefficient and wear, respectively.     

人工神经网络在塑性成形领域中的应用研究

在概述人工神经网络发展及应用的基础上,综述了国内外神经网络在塑性成形领域中应用的现状,着重介绍了板材成形和模具设计与制造方面应用神经网络的现状。利用人工神经网络实现了板材拉深成形智能化控制过程中摩擦系数的实时识别,并对其进行了实验证,结果表明,神经网络模型能够很好地解决板材拉深成形智能化控制过程中摩擦系数的实时识别问题,实现整个拉深化过程的智能化控制。最后指出了当前人工神经网络研究与应用中的不足之处及今后的研究方向。     

ANN在焊接接头抗弯强度预测中的应用

采用Ni—Fe—C合金作为填充金属,获得了基于11G焊的WC-30Co／45钢焊接接头。采用人工神经网络(ANN)方法,对WC-30Co／Ni-Fe-C／45钢11G焊过程中输入参数(焊接参数和填充金属成分)和力学抗弯强度之间的关系进行预测和分析。训练数据经过数据标准化处理,送入基于反向传播的多层前馈神经网络模型训练。并采用均方误差对模型进行误差分析。并采用训练的网络对焊接参数和填充金属成分与抗弯强度之间的关系进行预测。最后通过试验对预测结果进行了误差分析。结果表明,当采用碳含量(质量分数)0．6％或0．8％;Ni／Fe比为1．9～2．7的合金作为填充金属时可以获得较高的抗弯强度;构建的基于反向传播算法的ANN模型适用于评价WC-30Co／45钢TIG焊接头的抗弯强度,优于传统方法。     

Prediction of Tensile Property of Hydrogenated Ti600 Titanium Alloy Using Artificial Neural Network

An artificial neural network (ANN) model has been developed to analyze and predict the correlation between tensile property and hydrogenation temperature and hydrogen content of hydrogenated Ti600 titanium alloy. The input parameters of the neural network model are hydrogenation temperature and hydrogen content. The output is ultimate tensile strength. The accuracy of ANN model was tested by the testing data samples. The prediction capability of ANN model was compared with the multiple linear regression approach and response surface method. The combined influence of inputs on the tensile property is also simulated using ANN model. It is found that excellent performance of the ANN model was achieved, and the results showed good agreement with experimental data. Moreover, the developed ANN model can be used as a tool to control the tensile property of titanium alloys.     

The inverse of material properties of functionally graded pipes using the dispersion of guided waves and an artificial neural network

Using guided circumferential wave dispersion characteristics, an inverse method based on artificial neural network (ANN) is presented to determine the material properties of functionally graded material (FGM) pipes. The group velocities of lowest modes at six lower frequencies are used as the inputs of the ANN model. The distribution function of the volume fraction of the FGM pipe is fitted to a polynomial, then the outputs of the ANN are the coefficients of the fitting polynomial. The Legendre polynomial method is employed as the forward solver to calculate the dispersion curves for the FGM pipe. Levenberg–Marquardt algorithm is used as numerical optimization to speed up the training process of the ANN model.     

Prediction of martensite and austenite start temperatures of the Fe-based shape memory alloys by artificial neural networks

In this study, martensite start (Ms) and austenite start (As) temperatures of Fe-based shape memory alloys (SMAs) were predicted by using a back-propagation artificial neural network (ANN) that uses gradient descent learning algorithm. An ANN model is built, trained and tested using the test data of 85 Fe-based SMAs available in literature. The input parameters of the ANN model are weight percentages of seven elements (Fe, Mn, Si, Ni, Cr, Cu and Al) and three different treatment conditions (hot rolling, homogenizing temperature and quenching). The ANN model was found to predict the Ms and As temperature well in the range of input parameters considered. A computer program was devised in MATLAB and different ANN models were constructed with this program for prediction of As and Ms temperatures of iron-based SMAs. A comprehensive analysis of the prediction errors of Ms and As temperatures made by the ANN is presented. This study demonstrate that ANN is very efficient for predicting the Ms and As temperatures of iron-based SMAs.     

Prediction and analysis of the aging properties of rapidly solidified Cu-Cr-Sn-Zn alloy through neural network

It is known that the strength of a metal can be successfully improved by rapid solidification. The hardness of the rapidly solidified Cu-Cr-Sn-Zn alloy is much higher than that of the solution heat-treated and aged alloy. In this study, multiple-layer, feed-forward, artificial neural network (ANN) modeling has been used to study the hardness and electrical conductivity behavior of a rapidly solidified Cu-Cr-Sn-Zn alloy. The ANN model shows how the aging parameters influence the hardness and electrical conductivity of a rapidly solidified Cu-Cr-Sn-Zn alloy. The ANN modeling also provides encouraging predictions for information not included in the trained set samples, indicating that a backpropagation network is a very useful and accurate tool for property analysis and prediction.     

铸钢件耐磨堆焊层人工神经网络成分预测

阐述了人工神经网络在耐磨铸钢堆焊成分预测中的意义和工作原理，用自行编制的人工神经网络程序计算了铸钢件堆焊层的化学成分。     

人工神经网络在拉深系数选取中的应用

应用人工神经网络技术建立了多次拉深工艺中相对厚度值与极限拉深系数之间的BP网络模型 ,并用C语言编制了应用软件 ,将程序运行的结果与资料数据进行了比较和分析 ,表明人工神经网络应用于拉深系数方面是可行的     

基于人工神经网络的焊缝宽度预测

研究了用神经网络预测焊缝宽度的方法。首先对焊接质量检测系统的一些相关问题进行了研究,考虑了弧焊特性的提取、焊接质量的预报以及人工神经网络模型(ANN)的应用,设计了一个基于人工神经网络的焊接质量检测系统,给出系统的组成结构,ANN被用于预测焊缝宽度,建立了焊缝宽度预测的人工神经网络模型。为了验证建立的ANN模型的可行性,进行了仿真研究。仿真结果表明,所建立的ANN模型可预测焊缝宽度,基于人工神经网络的焊接质量检测系统是有效的。     

基于遗传算法的Al-5%Cu合金电脉冲孕育处理参数优化

以实验为基础,利用神经网络和遗传算法优化Al-5%Cu合金的电脉冲孕育处理工艺参数。神经网络的输入参数为脉冲电压、脉冲时间和电脉冲孕育处理时熔体温度,输出参数是合金凝固组织的晶粒度。在神经网络训练的基础上,采用遗传算法优化神经网络的输入参数。结果表明,神经网络和遗传算法的组合建模获得了较好的优化结果。     

Determining the stress intensity factor of a material with an artificial neural network from acoustic emission measurements

An artificial neural (ANN) network was trained to recognize the stress intensity factor in the interval from microcrack to fracture from acoustic emission (AE) measurements on compact tension specimens. The specimens were made from structural steel SWS490B whilst the ANN had a 5-14-1 structure. The number of neurons in the input layers was five inputs of the AE parameters such as ring-down counts, rise time, energy, event duration and peak amplitude. The performance of the ANN was tested using a specific set of the AE data. The ANN is a promising tool for predicting the stress intensity factor of material using AE data.     

肝移植患者他克莫司个体化给药研究

目的: 建立人工神经网络用于估算他克莫司血药浓度。 方法: 收集26例肝移植患者口服他克莫司的94份全血浓度数据,采用遗传算法配合动量法优化网络参数,建立人工神经网络。 结果: 人工神经网络平均预测误差(MPE)与平均绝对预测误差(MAE)分别为(-0.11±2.81) ng/mL 和(2.14±1.72) ng/mL,78.6%血药浓度数据绝对预测误差≤3.0 ng/mL。多元线性回归MPE与MAE分别为(0.56±2.70) ng/mL 和(2.15±1.63) ng/mL,9例次(9/14,64.3%)绝对预测误差≤3.0 ng/mL。人工神经网络准确性及精密度优于多元线性回归。 结论: 人工神经网络预测可用于预测他克莫司血药浓度,指导个体化给药。     

Modeling Blanking Process Using Multiple Regression Analysis and Artificial Neural Networks

The design of blanking processes requires the availability of a procedure able to deal with both tooling and mechanical properties of the workpiece material (blank thickness, hardness, ductility, etc.). This research presents the development and comparison of two models to predict the quality of the blanked edge represented by burrs height, the first model is an artificial neural network (ANN) based, while the second model is a multiple regression analysis (MRA) based. Finite Element modeling of the blanking process was used to generate the data for both models. Both ANN and MRA are able to give good prediction results, however, ANN still more accurate because it deals efficiently with hidden nonlinear relations when compared to MRA. The comparison between experimental and model results shows that average absolute relative error in the case of ANN was <2.20% for carbon steel and 4.85% for corrosion-resistant steel (CRES) compared to 15.18% for carbon steel and 14.22% for CRES obtained from the second order MRA. Therefore, by using ANN outputs, satisfactory results can be estimated rather than measured and hence reduce testing time and cost.     

Neural network detection of grinding burn from acoustic emission

An artificial neural network (ANN) approach is proposed for the detection of workpiece “burn”, the undesirable change in metallurgical properties of the material produced by overly aggressive or otherwise inappropriate grinding. The grinding acoustic emission (AE) signals for 52100 bearing steel were collected and digested to extract feature vectors that appear to be suitable for ANN processing. Two feature vectors are represented: one concerning band power, kurtosis and skew; and the other autoregressive (AR) coefficients. The result (burn or no-burn) of the signals was identified on the basis of hardness and profile tests after grinding. The trained neural network works remarkably well for burn detection. Other signal-processing approaches are also discussed, and among them the constant false-alarm rate (CFAR) power law and the mean-value deviance (MVD) prove useful.     

人工神经网络在钛合金领域的应用

概括了人工神经网络的学习机理以及运用较多的BP神经网络模型的BP算法原理,进一步综述了人工神经网络在钛合金材料高温变形行为研究、力学性能预测和相变规律等方面的应用情况。认为将人工神经网络技术应用于钛合金材料领域中,可以明显地提高工艺设计效率,缩短实验周期,对钛合金材料的研究具有很高的应用价值和深远的指导意义。